February CET

[Weather-history]

I find this surprising, aside from the idea of doing it to 2 decimal places but before I start another discussion/argument, its this

Hadley has just come out with a Feb CET of 2.8 (2.75 rounded up)

That makes this winter's CET 2.4 and 59th coldest ever

Now I was always taught in UK Met that numbers were ALWAYS rounded to the odd, ie 1.5C=1C in whole numbers thus 2.75C becomes 2.7C rounded to 1 decimal place.

Have they changed their own rules?

John when I did Maths at school I was always taught to round to the nearest even number when it comes to a .5

Edited March 4, 2010 by Mr_Data

[stewfox]

John when I did Maths at school I was always taught to round to the nearest even number when it comes to a .5

So 2.5 becomes 4 ??

2.5 would be rounded to 3 ?

2.4 rounded to 2

[Weather-history]

So 2.5 becomes 4 ??

The nearest even whole number to 2.5 is 2 :lol:

Edited March 4, 2010 by Mr_Data

[phil nw.]

The nearest even whole number to 2.5 is 2

It`s long time ago for me but i seem to remember being taught that any whole number followed by .5 was rounded up to the next whole number,ie 2.5 became 3 if rounding to whole numbers.

This was the rule i used up to GCE O level anyway.

I can never recall differentiating between odd and even numbers on this,unless this rule applies in Meteorology only.

Interesting this,never come across this before in my 61years.

[Weather-history]

I can never recall differentiating between odd and even numbers on this,unless this rule applies in Meteorology only.

No, it is a mathematical form of rounding called Gaussian rounding. Its a rounding that statisticians and bookkeepers are more likely to use.

Edited March 4, 2010 by Mr_Data

[phil nw.]

No, it is a mathematical form of rounding called Gaussian rounding. Its a rounding that statisticians and bookkeepers are more likely to use.

[

Thanks Mr.D for enlightening me.

I googled this and found an extended explanation on Wiki.

http://en.wikipedia.org/wiki/Rounding

in particular this para.

Quote,

Round half to even

A tie-breaking rule that is even less biased is round half to even, namely

• If the fraction of y is 0.5, then q is the even integer nearest to y.

Thus, for example, +23.5 becomes +24, +22.5 becomes +22, -22.5 becomes -22, and -23.5 becomes -24. This variant of the round-to-nearest method is also called unbiased rounding, convergent rounding, statistician's rounding, Dutch rounding, Gaussian rounding, or bankers' rounding. For most reasonable distributions of y values, the expected (average) value of the rounded numbers is essentially the same as that of the original numbers, even if the latter are all positive (or all negative). This is widely used in bookkeeping.""

Something new i have learned anyway.

Edited March 4, 2010 by phil n.warks.

[AderynCoch]

Is that the method the Met Office use? I ask because I have inserted daily CETs into a spreadsheet and sometimes the average doesn't quite match the official released figure.

I can't say I'm at all familiar with rounding .5 to the nearest even number (I was always taught to round .5 up to the next number, odd or even). What advantage does Gaussian rounding have?

[Paul T]

Is that the method the Met Office use? I ask because I have inserted daily CETs into a spreadsheet and sometimes the average doesn't quite match the official released figure.

I can't say I'm at all familiar with rounding .5 to the nearest even number (I was always taught to round .5 up to the next number, odd or even). What advantage does Gaussian rounding have?

The Met office round up, as its warmer - i.e; if March's cet is 5.51, they would round that to 6

[johnholmes]

No matter what technical reasons are given Met and WMO round to the odd number in temperature so 13.5C=13C and 14.5C=15C.

Only in pressure is it rounded down-for safety reasons in aviation thus 1014.5mb is given as 1014mb.

That is the way it was done in the nearly 40 years I worked in UK Met. Hence my querying of the initial value Mr D quoted.

I'll do a check on their site to see if they and WMO have changed the rules they previously operated under. Maybe my couple of links I still have into Met can enlighten me?

[johnholmes]

Hopefully this exchange of e mails will finally prove what I've tried to explain over the last two years to the doubters on Net Weather.

Dear John,

Thank you for your email. In reference to your enquiry, you are absolutely correct in what you have stated in your email, temperatures are rounded to the nearest odd number and pressure is rounded down. I hope this helps.

Kind Regards,

Tim Hill Weather Desk Advisor

Met Office, FitzRoy Road, Exeter, Devon, EX1 3PB, United Kingdom.

Tel: 0870 900 0100 or +44 (0)1392 88 5680 Fax: 0870 900 5050 Email: enquiries@metoffice.gov.uk http://www.metoffice.gov.uk

----------------------- Original Message ----------------------- From: John Holmes <john-holmes@zen.co.uk>To: "<enquiries@metoffice.gov.uk>"Date: Fri, 5 Mar 2010 07:38:00 +0000Subject: rounding temperatures and pressure

hi

Can you tell me please do UK Met and WMO still round temperatures to the nearest ODD number such 3.5C becomes 3C in whole numbers? 2.75C becomes 2.7C to one decimal place etc?Likewise in pressure is 1013.5mb rounded DOWN to 1013mb?

many thanks

John Holmes

and a final note re doing temperatures to 2 decimal places-another jh favourite rant!

From R Met S Weather magazine February 2010

'...whilst the practice of giving means to two decimal places when temperatures are only read to one decimal place is arithmetically unsound'.

Edited March 5, 2010 by johnholmes

[kold weather]

Now this is quite interesting, have the Met office rounded up for once to avoid a certain very rare cold statistic happening??

2.7C in the benchmark needed to get all three winter months 1.5C below the 30 year mean at the time, which IMO would be quite a feat and something that is rare (I'm not sure how many of those there are, 16-17 and 62-63 are two, are there any more?

I do wonder whether the Met office have gone against usual convention and rounded up to prevent that stat from happening, though that is probably the skeptical view!

[Thundery wintry showers]

Another possibility is that they might have been influenced by Philip Eden's figure being 0.1C higher still at 2.9. I remember back in 2005, Hadley regularly came out 0.2-0.4C lower than Philip Eden's figures and in the end they got revised upwards by approximately that amount. I can only be guessing though.

Alternatively the real figure might be something like 2.7525253636 and so they might have rounded it up to 2.8 because of the 3rd decimal point? Again though some question marks over that reasoning considering that even 2 d.p. is stretching viability given that temperatures are only measured to one tenth of a degree.

[johnholmes]

Now this is quite interesting, have the Met office rounded up for once to avoid a certain very rare cold statistic happening??

2.7C in the benchmark needed to get all three winter months 1.5C below the 30 year mean at the time, which IMO would be quite a feat and something that is rare (I'm not sure how many of those there are, 16-17 and 62-63 are two, are there any more?

I do wonder whether the Met office have gone against usual convention and rounded up to prevent that stat from happening, though that is probably the skeptical view!

Yes that is my next question to a link I have in there and to ask him if he can discover if they have they done that, if so, why?

Another possibility is that they might have been influenced by Philip Eden's figure being 0.1C higher still at 2.9. I remember back in 2005, Hadley regularly came out 0.2-0.4C lower than Philip Eden's figures and in the end they got revised upwards by approximately that amount. I can only be guessing though.

Alternatively the real figure might be something like 2.7525253636 and so they might have rounded it up to 2.8 because of the 3rd decimal point? Again though some question marks over that reasoning considering that even 2 d.p. is stretching viability given that temperatures are only measured to one tenth of a degree.

Please read my comment about more than 1 decimal point-2 they may do but beyond absolutely not-take it from me it just would not be taken past 2 decimal points.

[Thundery wintry showers]

Thanks for raising those questions with them- I will certainly be interested to know why (if they're prepared to divulge that info!) because it has me stumped at the moment.

[Weather-history]

Why the Met office don't use the Gaussian rounding is beyond me, it obvious that with a large data set such as the CET that the "bias" is smoothed out with the Gaussian method whilst with the method that they use there is a "bias" i.e figures are all rounded up.

[johnholmes]

Its an international agreement Kevin, all participating met services use the same system, and that applies to current data or past-CET type data, temp numbers always thrown to the odd-that is apart from the query raised about 2.75C=2.8C. I've not got round to asking about that yet.

[AderynCoch]

Interestingly, if you add up all the daily CETS for February (given to 1 d.p.) and divide by 28 you get 2.75 exactly. It's as cold a 2.8C month as you can get - downgrade just one of the daily values by 0.1C and you end up with 2.7C.